UUID - Added support for Oracle, redshift

Author: erezsh

data_diff/databases/oracle.py

@@ -16,6 +16,11 @@ class Oracle(ThreadedDatabase):
     TYPE_CLASSES: Dict[str, type] = {
         "NUMBER": Decimal,
         "FLOAT": Float,
+        # Text
+        "CHAR": Text,
+        "NCHAR": Text,
+        "NVARCHAR2": Text,
+        "VARCHAR2": Text,
     }
     ROUNDS_ON_PREC_LOSS = True
 
@@ -97,3 +102,7 @@ def offset_limit(self, offset: Optional[int] = None, limit: Optional[int] = None
             raise NotImplementedError("No support for OFFSET in query")
 
         return f"FETCH NEXT {limit} ROWS ONLY"
+
+    def normalize_uuid(self, value: str, coltype: ColType_UUID) -> str:
+        # Cast is necessary for correct MD5 (trimming not enough)
+        return f"CAST(TRIM({value}) AS VARCHAR(36))"

tests/test_database_types.py

@@ -291,7 +291,9 @@ def __iter__(self):
             "numeric",
         ],
         "uuid": [
-            "$uuid"
+            "text",
+            "varchar(100)",
+            "char(100)",
         ],
     },
     db.Oracle: {
@@ -310,7 +312,10 @@ def __iter__(self):
             "double precision",
         ],
         "uuid": [
-            "$uuid"
+            "CHAR(100)",
+            "VARCHAR(100)",
+            "NCHAR(100)",
+            "NVARCHAR2(100)",
         ],
     },
     db.Presto: {
@@ -407,8 +412,10 @@ def _insert_to_table(conn, table, values):
         for j, sample in values:
             if isinstance(sample, (float, Decimal, int)):
                 value = str(sample)
-            else:
+            elif isinstance(sample, datetime):
                 value = f"timestamp '{sample}'"
+            else:
+                value = f"'{sample}'"
             selects.append(f"SELECT {j}, {value} FROM dual")
         insertion_query += " UNION ALL ".join(selects)
     else: